1384.  The other part of the apparatus consisted of two insulating pillars, h and i, to which were fixed two brass balls, and through these passed two sliding rods, k and m, terminated at each end by brass balls; n is the end of an insulated conductor, which could be rendered either positive or negative from an electrical machine; o and p are wires connecting it with the two parts previously described, and q is a wire which, connecting the two opposite sides of the collateral arrangements, also communicates with a good discharging train r (292.).

1385.  It is evident that the discharge from the machine electricity may pass either between s and l, or S and L. The regulation adopted in the first experiments was to keep s and l with their distance unchanged, but to introduce first one gas and then another into the vessel a, and then balance the discharge at the one place against that at the other; for by making the interval at a sufficiently small, all the discharge would pass there, or making it sufficiently large it would all occur at the interval v in the receiver.  On principle it seemed evident, that in this way the varying interval u might be taken as a measure, or rather indication of the resistance to discharge through the gas at the constant interval v.  The following are the constant dimensions.

Ball s        0.93 of an inch. 
Ball S          0.96 of an inch. 
Ball l        2.02 of an inch. 
Ball L          0.62 of an inch. 
Interval v    0.62 of an inch.

1386.  On proceeding to experiment it was found that when air or any gas was in the receiver a, the interval u was not a fixed one; it might be altered through a certain range of distance, and yet sparks pass either there or at v in the receiver.  The extremes were therefore noted, i.e. the greatest distance short of that at which the discharge always took place at v in the gas, and the least distance short of that at which it always took place at u in the air.  Thus, with air in the receiver, the extremes at u were 0.56 and 0.79 of an inch, the range of 0.23 between these distances including intervals at which sparks passed occasionally either at one place or the other.

1387.  The small balls s and S could be rendered either positive or negative from the machine, and as gases were expected and were found to differ from each other in relation to this change (1399.), the results obtained under these differences of charge were also noted.

1388.  The following is a Table of results; the gas named is that in the vessel a.  The smallest, greatest, and mean interval at u in air is expressed in parts of an inch, the interval v being constantly 0.62 of an inch.